Upload
phungthuan
View
232
Download
0
Embed Size (px)
Citation preview
Modeling Centralized Organization of Organizational Change
Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
1Vrije Universiteit Amsterdam, Department of Artificial Intelligence
De Boelelaan 1081a, 1081 HV Amsterdam, The Netherlands
Email: {mhoogen, schut, treur}@cs.vu.nl
URL: http://www.cs.vu.nl/~{mhoogen, schut, treur} 2Radboud University Nijmegen, Nijmegen Institute for Cognition and Information
Montessorilaan 3, 6525 HR Nijmegen, The Netherlands
Email: [email protected]
URL: http://www.nici.ru.nl/~catholj
Abstract. Organizations change with the dynamics of the world. To enable organizations to change, certain
structures and capabilities are needed. As all processes, a change process has an organization of its own. In
this paper it is shown how within a formal organization modeling approach also organizational change
processes can be modeled. A generic organization model (covering both organization structure and behavior)
for organizational change is presented and formally evaluated for a case study. This model takes into account
different phases in a change process considered in Organization Theory literature, such as unfreezing,
movement and refreezing. Moreover, at the level of individuals, the internal beliefs and their changes are
incorporated in the model. In addition, an internal mental model for (reflective) reasoning about expected role
behavior is included in the organization model.
Keywords: organizational change, formal organizational modeling, organizational simulation,
multi-agent organizations, organization verification
1 Introduction
Within the literature on Organization Theory changing organizations play a dominant role
(Robbins 1998; Hyczynski and Buchanan 2001; Jaffee 2001). As change processes involve many
2 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
factors ranging from making the employees aware of changes to come and taking away resistance
to change to the design of efficient organizational structures. Changes can concern rather simple
processes of slight changes in one or more role descriptions. They may affect only a part of the
organization or practically the whole organization. Roles or big parts of the organization may be
deleted, new ones created. The realization of the organization probably changes, e.g., agents
fulfilling other roles than before, agents leaving the organization, agents joining the organization
(Glaser and Morignot 1997). A change may be initiated by the environment or by the organization
itself. The organization of a change process may involve agents from outside the organization
(e.g., consultation) or from inside. In this paper, the process of (business) organizational change is
analyzed in more detail. Methods used in this analysis are those of formalization, simulation and
verification. To organize change processes, a generic organization model for organizational change
is introduced and formalized. This organization model incorporates both multi-agent co-operation
aspects and individual cognitive aspects in the form of the internal mental states (e.g., beliefs) of
those involved in the change.
A specific area in which organizational change is inherent, is in the organization that is needed
to cope with a big upcoming event. Such an event can be a planned event in the area of sports or
concerts, for example, but also an incident that can grow out to a disaster. The latter area is the
focus of the project CIM (for Cybernetic Incident Management); cf. (Abbink et al., 2004;
Hoogendoorn et al., 2004; Hoogendoorn et al., 2005). A common characteristic for incidents and
big planned events is that the organizational structures start almost at zero, i.e., no activity, and
hence no organization, but (have to) grow out to a scale and form of organization that is able to
address large and complex processes by multiple parties and multiple agents. To test ideas on
organizational change modeling and to get more insight in cases with these characteristics, the
organization of a big sports event has been chosen: the famous Dutch 11 cities ice skating tour
(10.000s of people all performing 200 km of ice skating on one day, going from city to city). In
this case study the usefulness of the developed organization model for organizational change is
evaluated.
To model the organizational change process, the theory presented in (Lewin, 1951; Robbins,
1998) has been used as inspiration, and has been evaluated on its usefulness in an operational
(modeling) sense. The three phases unfreezing, moving, refreezing distinguished have been
incorporated in the generic model for organizational change developed. The case study shows that
Modeling Centralized Organization of Organizational Change 3
this theory indeed can be integrated in an organizational change modeling approach in a useful
manner.
This paper is organized as follows: Section 2 gives an overview of organizational change
literature, and introduces the stages that can be identified in an organizational change process.
Section 3 introduces the approach which has been used to model the stages in organizational
change in a formal way. The model itself is specified in Section 4 both from a structural as well as
from a behavioral perspective. Section 5 presents a language used to specify an organizational
model or an organization and Section 6 presents results of a case study which has been performed
to show how the approach can be applied. In Section 7 formal verification is performed upon the
simulation results to show that the simulated organization indeed satisfies the desired properties.
Finally, Section 8 draws conclusions based on the results presented in this paper.
2 Organizational Change Literature
Organizational change is a well studied topic in recent literature on sociology, psychology, and
economics. Change within organizations has become part of everyday life, some organizations are
even continuously undergoing change. Changing an organization is not a simple process, often
difficulties are encountered within such a change process. Research has shown that over 70 percent
of the change programs in organizations do not achieve the intended goal (Hall et al. 1993;
Bashein et al., 1994). Boonstra (2004) criticizes typical explanations given for these failures in that
they pay insufficient attention to the complexity of the change process itself. Three types of
organizational change are distinguished within the introduction of his book: First, planned
organizational change, which addresses questions with respect to problems that require change in
technical and instrumental aspects in which the problems and solutions are known. Secondly,
organizational development which is said to be suitable when “ the changes to be made are far-
reaching, the problems not entirely unambiguous but still recognizable, and there is some idea as
to the direction in which the solutions must be sought” . Cummings and Worley (20001) define
organizational development as “a system-wide process of applying behavioral science knowledge
to the planned change and development of strategies, design components, and processes that
enable organizations to be effective” . The final type of organizational change distinguished is
transformational change, in which the change processes include “renewal processes involving
4 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
actors from various organizations” . In Ackerman (1986) transformational change is said to be the
emergence of a totally new state of being out of the remains of the old state.
Both in planned change and organizational development an approach is taken in which a move
is performed form one stable state to another. The change processes involve the phases in which an
organization is unfrozen, changed, and refrozen. These phases within the organizational change
process originate from the ideas of Kurt Lewin (1951). He states that there are two opposing forces
at work when changing an organization: forces that resist the change, and forces that drive towards
the newly desired organization. Figure 1 presents the phases and forces within organizational
change in a graphical manner (from Robbins, 1998). The unfreezing phase begins at the moment
that change becomes necessary and consists of the process of changing the resisting and driving
forces in such a way that change becomes possible (i.e., the driving forces outweigh the resisting
forces). Both Schein (1993) and Hosking (1999) stress the importance of communication within
this unfreezing phase to enable a successful change. According to Cummings (2004)
organizational development has discovered a long list of causes for resistance to change, such as
structural inertia, work habits, fear of the unknown, powerful interests, and members’ security
needs. Forces that drive an organization to change can be found in Jaffee (2001) and for example
include change on the supply side, customer behavior, available technology (see e.g. Orlikowski
and Hofman, 1997), etc. The actual change of the organization is contained in the movement phase
in which the organization is moved from the current state to the desired stated. The refreezing
phase involves freezing the newly formed organization so that there is no possibility to return to
the former status quo or to continue changing in another unwanted direction. The whole re-
Fig. 1. Movement of an organization from a status quo to a desired state (Robbins 1998)
Unfreezing Movement Refreezing
Resistant forces
Driving Forces
Time
Desired state
Status quo
Modeling Centralized Organization of Organizational Change 5
organization process is completed when all phases have been completed. The unfreezing can be
performed by increasing the driving forces and/or by decreasing the resisting forces. In his book,
Cummings and Worley (2001) states that Lewin’s model remains closely identified with the fields
of planned change and organizational development.
Since the model of Lewin is a highly generic model, effort in organizational development
research has gone into making it more concrete. Lippitt et al. for example arrange Lewin’s model
in seven steps: within the unfreezing phase they identify scouting, entry and diagnosis. The
movement phase is split up into planning and action, and finally, stabilization and evaluation, and
termination are placed within the refreezing phase.
Particularly of interest for this paper are further refinements regarding the actors within
organizational change. Kotter (1998) has defined characteristics for change managers to prevent
organizations from falling into pitfalls due to bad change management. These include having
industrial and organizational knowledge, relations in the firm and industry, and reputation and
track record. Power is an important aspect related to actors in organizational change processes as
well, since the resisting and driving forces of the actors need to be changed to enable an
organizational change. This particular research branch is called power dynamics. Research started
in 1946 when Kurt Lewin introduced T-groups in a laboratory training setting and was mainly
based on group-based approaches where people learn about group dynamics, leadership and
interpersonal relationships. Bradshaw and Boonstra (2004) identify several different notions of
power. Firstly, manifest-personal power which takes the viewpoint that a person can have power
over other people and can make them do something they would not do otherwise. Research
concerning this form of power research is said to have started with the work of Dahl (1975),
Emerson (1962) and Wrong (1968). In manifest-structural power, power is no longer viewed from
the personal perspective, but from a group perspective. Bacharach and Lawler (1980) is named as
a reference for this notion of power. Negotiations are said to be an important part of the models
regarding manifest structural power. Latent-Cultural Power sees organizing as “a process of the
creation and reproduction of shared meanings that are largely latent or unconscious” , they also
refer to Alvesson (1993) for more details about the notion of latent-cultural power. Finally, latent-
personal power which is said to be relatively new in organization theory. This type of power is
said to differ from latent-cultural power is several different ways. First of all, power is said to be
scattered throughout the organization, even individuals at the bottom of the organization can
6 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
deploy their power. Secondly, power relations are assumed to become part of the psyche of the
individual.
As the theory of Lewin is still considered being the underlying theory for organizational
change research and considered valid, this paper tries to model the theory in a generic sense as a
first step towards modeling and understanding complex organizational change processes. Further
extensions might focus on the idea sketched above such as on more complex power relationships,
the role of different characteristics for change managers, the different ways to enable unfreezing an
organization.
3 Modeling Approach for Organizations
Before being able to model the organizational change processes identified by Lewin, a
methodology is required which enables modeling organizations in general. This section presents
such a methodology which allows modeling of organizations from two perspectives. First, the
structural perspective, merely specifying the structural blueprint of an organization, and secondly,
the behavioral perspective which specifies the behavioral of an organization and the actors within
such an organization.
3.1 The structural description of an organization
For the structural description of actual multi-agent organizations, the AGR (for agent/group/role)
model has been adopted (Ferber and Gutknecht, 1998). In that approach, an organization is viewed
as a framework for activity and interaction through the definition of groups, roles and their
relationships. But, by avoiding an agent-oriented viewpoint, an organization is regarded as a
structural relationship between a collection of agents. Thus, an organization can be described
solely on the basis of its structure, i.e. by the way groups and roles are arranged to form a whole,
without being concerned with the way agents actually behave, and multi-agent systems will be
analyzed from the outside, as a set of interaction modes. The specific architecture of agents is
purposely not addressed in the organizational model. The three primitive definitions are:
• The agents. The model places no constraints on the internal architecture of agents. An agent is
Modeling Centralized Organization of Organizational Change 7
only specified as an active communicating entity which plays roles within groups. This agent
definition is intentionally general to allow agent designers to adopt the most accurate definition of
agent-hood relative to their application.
• Groups are defined as atomic sets of agent aggregation. Each agent is part of one or more
groups. In its most basic form, the group is only a way to tag a set of agents. An agent can be a
member of n groups at the same time. A major point of these groups is that they can freely overlap.
• A role is an abstract representation of an agent function, service or identification within a group.
Each agent can handle multiple roles, and each role handled by an agent is local to a group.
Figure 2 presents an example of an organization modeled in AGR. The large ovals denote
groups whereas the smaller ovals denote the roles within the organizations. Furthermore, the
solid arrows denote interactions between roles, and the dashed lines represent inter-group
interactions. Agents realizing the roles are not depicted.
To enable simulation and reasoning about such an organizational model, the Structural
Language SL is used, based on the set of sorts (a class or type of objects) that is shown in
Table 1. These sorts enable talking about structural elements in the organization model.
Additionally, Table 2 shows a set of predicates within SL that define relations between the
introduced sorts.
����� �����
����� ����� ���� ����
���� �
���� � ���� �
Fig. 2. Example organization modeled within AGR
8 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
Table 1. Sorts in SL
Sort Description
ROLE Sort for a role within an organization.
AGENT Sort for an agent that can be allocated to a certain role.
GROUP Sort for a group within an organization.
TRANSFER Sort for a connection between two roles within one group.
GROUP_INTERACTION Sort for a connection between two roles in a different group.
Table 2. Predicates defined in SL to describe the structure of an organization
Predicate Description
exists_role: ROLE A role exists within an organization.
allocated_to: AGENT x ROLE x GROUP An agent is allocated to a role within a group.
exists_group: GROUP A group exists within the organization.
role_belongs_to_group: ROLE x GROUP A role belongs to a group.
intra_group_connection: ROLE x ROLE x GROUP x
TRANSFER
A role is connected to another role (directed)
within a certain group by means of a transfer
connection. The source and destination roles are
allowed to be equivalent.
inter_group_connection: ROLE x GROUP x ROLE x
GROUP x GROUP_INTERACTION
A role within a group is connected to a role within
another group by means of a group interaction
connection.
3.2 The behavioral description of an organization
In this section a method to express dynamics within an organizational model is addressed. To
formally specify dynamic properties at the different aggregation levels that are essential in an
organization, an expressive language is needed. To this end the Temporal Trace Language is used
as a tool; cf. (Jonker and Treur 2002). For the properties occurring in the paper informal, semi-
formal or formal representations are given. The formal representations are based on the Temporal
Trace Language (TTL), which is briefly described as follows; for more formal details, see
Appendix A.
Modeling Centralized Organization of Organizational Change 9
A state ontology Ont is a specification (in order-sorted logic) of a vocabulary. A state for
ontology Ont is defined as an indication of which state properties expressed in ontology Ont hold
in the state and which do not hold. The set of all states is modeled by the sort STATE. A fixed time
frame T is assumed which is linearly ordered. A trace or trajectory γ over a state ontology Ont
and time frame T is an indication of which state occurs at which time point, for example if a
discrete time frame based on natural numbers is taken, a trace is a sequence of states γt (t ∈ T). The
set of all traces over state ontology Ont is modeled by the sort TRACE. Depending on the
application, the time frame T may be dense (e.g., the real numbers), or discrete (e.g., the set of
integers or natural numbers or a finite initial segment of the natural numbers), or any other form,
as long as it has a linear ordering. A dynamic property over state ontology Ont is a temporal
statement that can be formulated with respect to traces based on the state ontology. Such temporal
statements can express, for example, a temporal relationship between the fact that in a given trace
a certain state property holds at a certain time point and another state property holds at some other
time point. For more formal details, see Appendix A.
The Temporal Trace language can be used to specify behavioral properties at different
aggregation levels, according to the organizational structure. Within the AGR approach the
aggregation levels are the level of the roles, the level of the groups and the level of the
organization as a whole (see Figure 3). The lower level properties can often be modeled in simpler
formats than the higher level properties. In particular, it is often possible to model the properties at
the leaves of the tree in the form of directly executable properties, i.e., by direct temporal
dependencies between state properties in two successive states. To model direct temporal
dependencies between two state properties, not the expressive language TTL, but the simpler leads
to format is used. This is an executable format that can be used to obtain a specification of a
simulation model in terms of local dynamic properties (the leaves of the tree in Fig. 3). The format
is defined as follows. Let α and β be conjunctions of elementary state properties, and e, f, g, h non-
Fig. 3. Overview of interlevel relations between dynamic properties
transfer properties role properties
group properties intergroup interaction properties
organization properties
10 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
negative real numbers. In the leads to language α →→e, f, g, h β, means:
if state property α holds for a certain time interval with duration g,
then after some delay (between e and f) state property β will hold
for a certain time interval of length h.
For a precise definition of the leads to format in terms of the language TTL, see (Jonker and Treur,
2002). A specification of dynamic properties in leads to format has as advantages that it is
executable and that simulation results can be depicted graphically.
Table 3 shows the predicates within the Behavioral Language BL which allows the specification
of the behavioral part of the organization at different aggregation levels, using the TTL language
as described above. The sort DYNPROP expresses an identifier of a dynamic property whereas
DYNPROPEXP expresses the dynamic property itself in terms of TTL.
Table 3. Predicates defined in BL to define the dynamics within an organization
Predicate Description
role_property: DYNPROP x ROLE x GROUP A role within a group has a role property.
transfer_property: DYNPROP x ROLE x ROLE x GROUP Within a group, a transfer property with an
identifier holds between two roles.
group_property: DYNPROP x GROUP A group has a certain group property.
group_interaction_property: DYNPROP x ROLE x GROUP x
ROLE x GROUP
An interaction property with an identifier holds
between two roles in different groups.
organization_property: DYNPROP A certain or property holds for the organization.
has_expression: DYNPROP x DYNPROPEXP A specific dynamic property has an expression.
Based on the sort DYNPROPEXP it is possible to put more constraints on particular types of
properties. The constraints for the different properties are defined below. The formal
representations of these properties can be found in Appendix B.
Role dynamic properties
Role properties involve only one role, namely the role for which the property holds. Therefore, a
role property should only contain elements that are part of the ontology of that role. The group is
also part of the definition of the ontology since roles in different groups can have the same name
and might have a different ontology. Role properties can be divided into different types which in
Modeling Centralized Organization of Organizational Change 11
turn can be defined more restricted than the general definition. An example of such a refinement is
an executable role dynamic property.
Transfer dynamic properties
Transfer properties relate the output of a role to the input of a destination role, therefore the
restriction on this dynamic property is that it should be expressed in terms of the output ontology
of the source role combined with the input ontology of the destination role.
Group dynamic properties
Group dynamic properties are dynamic properties expressed in terms of the state ontologies of
(some of) the roles within the group. The most common type of group property relates an output
state of a role within the group to an input state of another role within that group.
Intergroup interaction dynamic properties
Group interaction properties involve the input of a role within one group which is related to the
output of a role within another group.
Organization dynamic properties
For the organization dynamic properties the same holds as for group properties: states of
multiple roles (this time in different groups) can be involved; there is no further specific
definition for this type of property.
4 Organizing Organizational Change
The term organizing organizational change makes it explicit that organizational change is a
behavior process of that organization. Therefore, when formalizing organization dynamics, also
the process of change must be formally specified as one of the possible ways of behavior of the
organization. As all organizational behavior is described in terms of the behavior properties of the
roles in that organization, also the whole process of organizational change is attributed to a set of
roles in that organization. This section presents an organization model of organizational change
that is based on the three stages of change introduced by Lewin.
12 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
4.1 Structure and Informal Behavior of the Change Organization
Modeling the forces indicated in Lewin’s model entails attributing these forces to roles. Given an
existing organization model that does not model organizational change, there are two basic choices
that can be made: assigning these forces to roles already in the model, or extending the model with
additional organizational elements. The first can be a part of the second approach by first
extending the existing model with additional organizational elements, and then applying the first
approach. Although the first approach can be a part of the second, when modeling an organization
in which the realizing agents cannot reason about the change or even about the role that they are
playing (e.g., when modeling an ant hill), only the first approach can be followed and the roles
must be modeled as adaptive roles to ensure the possibility of change. In this article, the realizing
agents can reason about roles and organizations. The second approach is chosen to most explicitly
show the modeling process. In both cases the behavioral specification of the organization elements
needs extension, resulting in an organization model that incorporates organizing organizational
change.
Consider, as an example, the organization as presented in Section 3.1, Figure 2 which is also
shown at the bottom part of Figure 4a. An organizational change might for example concern the
removal of Group3, which in turn could imply that one of the agents realizing the organization will
be fired. It might further entail a re-allocation of agents over roles in groups. The organization in
Fig. 4. (a) Organization before the change (b) Organization after the
organizational change
Modeling Centralized Organization of Organizational Change 13
its state before change resists change (resisting forces outweigh the driving forces). To formally
model this phenomenon, the resisting and driving forces must be attributed to roles. Attributing
them to the existing roles is counterintuitive, because different roles have been identified to specify
different behaviors. The resisting and driving behaviors are of a different category. The way
chosen in this article, is to recognize that all agents part of the realization of the organization have
one thing in common: they are all members of the organization. Some members of the
organization might be in favor of change, some against, and this might change over time. This is
modeled by adding the role Member to the organization model, and attributing driving and
resisting forces to that role. Given that the organization changes from one stable situation to a new
stable situation, there is a need to model the focus existing in the organizational change. For this
reason the role of Change Manager is added to the organizational model. The Change Manager is
attributed with driving forces. This role can be realized by an agent from an external company, i.e.
a consultant type of role, or by an agent from within the organization. In Figure 4(a), the new roles
are grouped together in an organizational element called the Change Group, the members are
represented by Member One, Member Two, etc.
The Change Group is depicted in grey in Figure 4(a) to indicate that in stable situations this
group is inactive. The Change Manager can be of several different types, for example there can be
a global Change Manager, that is allowed to change the entire organization. It is however also
possible to have a local Change Manager that is only allowed to change a certain part within an
organization and therefore can only communicate with a sub-group of the members within the
Change Group. Because the Change Manager can be a representative of the company itself or of
an external company there is no predefined shared allocation between this role and another. Every
realizing agent of the organization is (next to the role it was already allocated to) also allocated to
one instance of the Member role of the Change Group. The Change Group has a meta-view on the
organization, and can, therefore, be seen as a meta-group. The start of an unfreezing phase
(meaning a change is due) is characterized by a sudden activity of the Change Manager within the
Change Group. The Change Manager might, for example, inform (all or some of) the instances of
the Member role of the impending organizational change and the reasons for this change. Aside
from the resulting reduction of resisting forces that this information might bring about, this
interaction can also be used to model the preparation for the movement phase.
At the end of a well-performed unfreezing stage, maybe all Member role instances, but at least
every Member role instance whose realizing agent is somehow involved in the change, now has
14 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
beliefs about which role its realizing agent may have to play in the new organization. These beliefs
include the expected role behavior. The end of the unfreezing phase may be characterized by the
presence of these beliefs in the respective member role instances or communication of this
presence to the Change Manager. Note that this does not say anything about all activities required
to accomplish these shared beliefs.
The start of the movement phase, after a well-performed unfreezing phase, is characterized by
the Change Manager informing all Members of when the actual change in organization is to take
place. At the indicated moment, all Member roles are to consider in their beliefs the new
organization form to be the current organization form. The movement phase is used to achieve (for
example, by being informed) that all involved will get the appropriate beliefs on the new structure
and their roles in this structure. As a result, the affected parts of the organization will start
behaving according to the behavior specification of the new organization form. This process is
modeled by means of the shared allocation of agents. Behavior that has become obsolete within
the organization will disappear over time .
The start of the refreezing phase is characterized by regular functioning of the new
organization form and a de-activation of the Change Group, see Figure 4(b). The refreezing phase
is complete when the behavior of the organization shows the routines that correspond to the
expected behavior of the new, now current, organization.
Next to the structural properties of the organization model of organizational change, also the
behavioral properties of the roles involved should be described to get a complete model. The next
sections describe the behavioral properties of the main roles; the Change Manager and the
Member.
4.2 Dynamic Properties for the Behavior of the Change Organization
The Change Manager is active in all stages of the organizational change. The properties in this
section are described in a domain independent manner, more describing the global behavior than
the actual behavior. Examples of more specific properties can be found in Section 6. First,
properties regarding the unfreezing phase are presented, after which the behavior during the
movement phase is described. Finally, the behavior during the refreezing phase is described.
Modeling Centralized Organization of Organizational Change 15
4.2.1 Dynamic properties for the Unfreezing Phase
First of all, the following property states the global behavior during the unfreezing phase, namely
that once there is an upcoming change, eventually enough key Members (fraction e) within the
Change Group will be unfrozen which takes the form of a communication of acceptance of the
new organization model.
GP1(ChangeGroup): Unfreezing Organization
if at time t the ChangeManager within the ChangeGroup has access to a plan which specifies a condition C for a
decision to reorganize based on a new organizational model OM
and condition C is met at time t,
and the Change Manager uses fraction e
then at a later point in timet2 , at least fraction e of the key Members within the ChangeGroup will have informed the
ChangeManager upon their acceptance of the new organization model OM.
This property can be fulfilled by means of several lower level properties. First of all, the Change
Manager informs all Member within the Change Group that are involved in the change on the new
organizational model.
RP1(ChangeManager): Communicate Change
if at time t the ChangeManager within the ChangeGroup has access to a plan which specifies a condition C for a
decision to reorganize based on a new organizational model OM
and condition C is met at time t
and Member M1 is involved in the change to organizational model OM at time t
then at a later point in time t2 the ChangeManager will inform Member M1 about the upcoming change to
organizational model OM.
Furthermore, ideally once a Member is informed about such an upcoming change, the member will
eventually communicate the acceptance of the new organizational model OM and thus will show
to be unfrozen. Fraction e is given as a parameter for the property. Note that these properties
describe a successful unfreezing phase where at least fraction e of the Members accepts the
change.
GP2(ChangeGroup, e): Confirm Change Acceptance
for at least a fraction e of the key Members in the Change Group
if at time t a Member M1 is informed about a new organizational model OM
then at a later point in time t2 Member M1 will inform the ChangeManager of its acceptance of the change to
new organizational model OM.
The property above is again specified in a general sense, as there might be a whole process
involved in convincing the Member of the improvements that come with the new organizational
model M. Hence, there are two ways in which property GP2 can be fulfilled by the Members.
16 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
First, the Member can immediately agree with the organizational model, and as a result be
unfrozen at once.
RP2(Member):No Resistance to Change
if a Member is informed about a new organizational model M at time t,
then at a later point in time t2 Member M1 will inform the ChangeManager upon its acceptance of the new
organizational model OM
and there does not exist a time t’ between t and t2 at which Member M1 has expressed resistance to the change to the
organizational model OM.
Another option is that a Member expresses temporal resistance to the change.
GP3(ChangeGroup): Belief Change after Resistance
for all Members in the Change Group
if a Member M1 is informed about a new organizational model OM at time t,
then at a later point in time t2 Member M1 will inform the ChangeManager of its resistance to the change to
the new organizational model OM,
and at a time t3 later than t2 Member M1 will inform the ChangeManager of its acceptance of the new
organizational model OM.
Opposition to change can be split up into several lower level properties. First, the Member
opposes the change.
RP3(Member): Oppose to Change
if a Member is informed about a change to a new organizational model OM at time t,
then at a later point in time t2 Member M1 will inform the ChangeManager of its resistance to the change to
the new organizational model OM.
In response the Change Manager puts forward a communication that hopefully will convince
the Member that organizational model OM is an appropriate option for him. Note that these
terms are kept abstract on purpose as there are many ways to convince such Members in
organizational change literature, and depending on the particular case a choice can be made
(see also Section 6).
RP4(ChangeManager): Convince Member
if Member M1 informs the ChangeManager of its resistance to the change to the new organizational model
OM at time t,
then at a later point in time t2 the ChangeManager will put forward additional arguments to Member M1 for
the change to the organizational model OM.
Once this information is received by the Member it is assumed that he will be unfrozen.
RP5(Member): Member Convinced
if Member M1 receives additional arguments for organizational model OM at time t,
and Member M1 is convinced by the additional arguments for organizational model OM at time t,
then at a later point in time t2 Member M1 will inform the ChangeManager upon its acceptance of the
Modeling Centralized Organization of Organizational Change 17
organizational model OM
There is also the possibility that a Member does not get convinced, which means that the Member
again communicates resistance. The Change Manager can put forward more arguments in response
(or use another method from organizational change theory). The possibility exists that not enough
key Members of the organization communicate the acceptance of the organizational model OM,
resulting in an organization which is not unfrozen. To show the relation between the different
properties for a successful unfreezing phase, Figure 5 shows a property tree.
The tree depends upon the number of Members involved in the change of the organization, this
tree covers an organization which only consists of one Member.
4.2.2 Dynamic properties for the Movement Phase
The movement phase is rather straightforward after the unfreezing phase, in case a fraction e of the
key Members have communicated their acceptance of the organizational change towards the
organizational model OM, and the condition for the change to occur holds, the roles within the
groups of the organization will show the behavior as specified in the organizational model OM.
Property OP1 specifies this movement and is referred to as an organizational property as it also
includes roles outside of the Change Group.
OP1: Successful move
if at time t the ChangeManager within the ChangeGroup has access to a plan which specifies a condition C for a
RP1: Communicate Change
GP1: Unfreezing organization
GP2: Confirm Believe Change
RP2: No Resistance to Change
GP3: Belief Change after Resistance
RP5: Member Convinced
RP4: Convince Member
RP3: Oppose to Change
Fig. 5. Unfreezing property hierarchy specified by means of an AND/OR tree
18 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
decision to reorganize based on a new organizational model OM
and condition C is met at time t,
and the Change Manager uses fraction e
and at least fraction e of the key Members involved in the change have informed the ChangeManager of their
acceptance of the change to organizational model OM at time t,
and organizational model OM specifies behavior B for a role R within group G at time t,
then at a later point in time t2 role R within group G behaves according to the behavior specification B.
Satisfaction of this high level property can be accomplished by means of a group property for the
Change Group and group interaction properties between the ChangeGroup and the other groups
within the organization. First, the group property states that all Members involved in the change
will receive an announcement of the organizational model being activated, as expressed in GP4
below.
GP4(ChangeGroup, e): Change Activation
if at time t the ChangeManager within the ChangeGroup has access to a plan which specifies a condition C for a
decision to reorganize based on a new organizational model OM
and condition C is met at time t,
and at least fraction e of the key Members involved in the change have informed the ChangeManager of their
acceptance of the organizational model OM at time t,
then at a later point in time t2 all Members involved in the change have received the announcement of organizational
model OM being active.
This property is entailed by two lower level properties. First, the ChangeManager announces the
activation of the of the organizational model OM based on the conditions specified.
RP6(ChangeManager): Announce Change
if at time t the ChangeManager within the ChangeGroup has access to a plan which specifies a condition C for a
decision to reorganize based on a new organizational model OM
and condition C is met at time t,
and all Members involved in the change have informed the ChangeManager of their acceptance of the
change to organizational model OM at time t,
then at a later point in time t2 the ChangeManager announces the new organizational model M being active.
And furthermore, this information is received by the Members via transfer property TP1.
TP1(Change Manager, Member): Transfer announcement
if at time t the ChangeManager announces organizational model M being active
and at time t Member M1 is involved in the change to organizational model M
then at a later point in time t2 Member M1 will receive this announcement on his input.
Finally, the group interaction properties state that after the announcement has been received by a
Member role, the roles with which the Member that receives the announcement of activation will
show the behavior as specified in the organizational model OM, expressed in GIP1.
Modeling Centralized Organization of Organizational Change 19
GIP1(Member, ChangeGroup, R, G): New organization active
if at time t Member M1 is informed about a new organizational model OM being active,
and Member M1 has a shared allocation with a role R within group G at time t,
and role R has a behavior description B in organizational model OM at time t,
then at a later point in time t2 role R within group G will start behaving according to behavior B.
Figure 6 shows the property hierarchy for the movement phase.
4.2.3 Dynamic properties for the Refreezing Phase
The final step in the model of Lewin entails refreezing the organization. Within the model
presented in this paper, this is expressed in the following way. There are two conditions to start the
refreezing phase. First, the organizational model OM has been activated. Second, all roles are
actually behaving according to the behavior specification. During the refreezing phase, key
Members inform the Change Manager about what roles are showing the correct behavior. In case
enough of these key Members (i.e. a fraction e1) communicate that a critical mass of roles (i.e.
fraction e2) indeed show the correct behavior for a sufficient period of time p2, after a
conditioning phase of length p1, the refreezing phase is said to be ended successfully. This
property is expressed as OP2.
OP2: Successful Refreezing
if before time t the ChangeManager has informed the Members that a new organizational model OM is active,
and at time t all the roles within organization are just behaving according to behavior specification B within the
organizational model OM,
and the Change Manager uses a conditioning period p1, a critical period of length p2, and fractions e1 and e2,
then there exists a time point t2 (t2 > t + p2) such that at t2 the Change Manager is informed by at least fraction e1 key
elements that behavior B is efficiently performed by at least fraction e2 of the roles within the organization over
OP1: Successful move
RP6: Announce change
GP4: Change activation
GIP1: New organization active
TP1: Transfer announcement
Fig. 6. Movement property hierarchy specified by means of an AND tree
20 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
the last period p2
The property can be accomplished by means of a group interaction property and a group property.
First, the group property states that from the time point the behavior is first shown by a role R
within group G, there exists a time point at which the role R has shown the correct behavior for the
minimum duration p, set by the Change Manager. The fraction e2 and periods p1 and p2 are
specified as parameters.
GP5(G, e2, p1, p2): Show Proper behavior
for at least a fraction e2 of the roles within group G,
if at time t a role R within group G just shows behavior B
then there exists a time point t2 (t2 > t + p1 + p2) such that at all time points between t2 and t2 – p2 role R within
group G shows behavior B
Some roles will immediately satisfy this property within the group, as specified by property RP7.
This means that the behavior shown is always according to the specified behavior.
RP7(R, p1, p2): Immediately Show Behavior
if at time t a role R within group G just shows behavior B
then for all time points t2 such that t < t2 � t + p1 + p2 t role R within group G shows behavior B
Of course it is also possible that the role R within group G falls back into its old habits, not
complying to the behavior specification within the new organizational model. After correction
however, the role shows the correct behavior again in case of successful refreezing. Such
temporarily falling back into old habits is specified in property GP6.
GP6(G, p1, p2): Show Behavior after Correction
for at least a fraction e2 of the roles within group G
if at time t a role R within group G just shows behavior B
then there exists a time point t1 > t and t1 < t + p1 at which role R within group G does not show behavior B
and there exists a time point t2 (t2 � t1 + p1 + p2) such that at all time points between t2 and t2 – p2 role R within
group G shows behavior B
Property GP6 is entailed by three lower level properties. First, RP8 expresses the improper
behavior of the role R:
RP8(R, p1, p2): Improper behavior
if at time t a role R within group G just shows behavior B
then there exists a time point t2 < t + p1 at which role R within group G does not show behavior B
To correct this improper behavior, another role within the same group can correct role R by
reminding the role of the proper behavior B:
RP9(R): Correct Improper Behavior
if at time t a role R within group G does not show the required behavior B
then at a later point in time another role R2 within group G will remind role R within group G of the proper
Modeling Centralized Organization of Organizational Change 21
behavior.
In a successful refreezing phase the correction indeed works, and role R returns to the correct
behavior again (RP10). In case the role R is not properly refrozen such a correction might
not work and therefore role R will continue to show the unwanted behavior.
RP10(R): Behave correct again
if at a time point t role R within group G is reminded by role R2 within group G of the proper behavior he
should show
then for all later points in time t2 > t role R within group G shows behavior B as long as no new reorganization has
been announced
Finally, GIP2 specifies that after having shown the correct behavior for a period longer than length
p, the Member within the Change Group communicates this to the Change Manager.
GIP2(R, G, Member, ChangeGroup, p1, p2): Communicate correct behavior
if between time point t and t2 (where t2 > t + p2) role R within group G shows the behavior according to B
and role R within group G has a shared allocation with Member M1
then at time t2 +1 Member M1 informs the Change Manager within the Change Group that behavior B is efficiently
performed by role R within group G over the last period p2.
The property hierarchy for the refreezing phase is shown in Figure 7.
GIP2: Communicate correct behavior
OP2: Successful Refreezing
GP5: Show proper behavior
RP7: Immediately Show Behaviour
GP6: Show Behavior after Correction
RP10: Behave Correct Again
RP9: Correct Improper Behavior
RP8: Improper Behaviour
Fig. 7. Refreezing property hierarchy specified by means of an AND/OR tree
22 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
5 Change Language
Since communication between the ChangeManager and the Members within the ChangeGroup
also concerns changes to the current organization (i.e., a new organizational model), this section
describes functions for describing such changes to be made. The sorts that have been used for this
language are shown in table 4, and are basically the sorts that make it possible to use the structural
and behavioral languages SL and BL introduced in Section 3. Moreover, the sort ACTION models
actions that can be performed. If a conjunction of elements of ORG_ELEMENT is deleted or
added, then all conjuncts are removed from or added to the model.
Table 4. Sorts used for the functions to describe organizational change.
Sort Description
ORG_BEHAVIOR_ELEMENT Defined by the behavioral language BL.
ORG_STRUCTURE_ELEMENT Defined by the structural language SL.
ORG_ELEMENT Union of the sorts ORG_BEHAVIOR_ELEMENT or and
ORG_STRUCTURE _ELEMENT
ORG_PART Conjunctions of elements from ORG_ELEMENT.
ACTION Sort for actions.
The functions and predicates that can be used to describe organizational change are shown in
table 5. The modify function is basically a combination of the delete and add function, but because
it is most likely that change includes modification of certain elements it is more intuitive to include
it as a function. The add function possibly takes a conjunction of ORG_ELEMENT as an input
(denoted as ORG_PART), this however is impossible for the delete because this would not result
in a unique system configuration. The performance of the actions is done internally within the role,
resulting in a communication that the structure is in place.
Modeling Centralized Organization of Organizational Change 23
Table 5. Functions and predicates used to describe organizational change.
Function or Predicate Description
add: ORG_PART � ACTION Add takes an ORG_PART and creates the
action to add that part.
delete: ORG_PART � ACTION Delete takes an ORG_ELEMENT and creates
the action to delete it.
modify: ORG_PART x ORG_PART � ACTION The first element models the current
organization, the second specifies the
modifications that need to be done. An action
is constructed by means of this.
to_be_performed: ACTION Predicate that a certain action is to be
performed. This can be add, delete or modify.
An organization model for organizational change as described informally in Sections 2 and 3,
involves a number of issues:
• changing internal (belief) states of all those involved in the changing organization
• changing organization structure
• taking up new roles by agents
• internal state properties of the agents involved incorporate beliefs on organization structure
as well as beliefs on dynamic properties characterizing role behavior
• internal state properties (beliefs) play a role as part of the dynamic properties characterizing
role behavior
A language to express dynamic properties of a changing organization has to be a rich language
able to express all these aspects in combination. Such a language is defined in this section as an
extension of TTL (Jonker and Treur 2002) called meta-TTL. Note that in this language not only
dynamic properties are defined on top of state properties, but also state properties (in particular
beliefs) are defined on top of dynamic properties. This makes it possible to express a dynamic
property built using a belief state property which itself refers to a dynamic property, and so on. So
on the top level this is a dynamic property built on state properties (the beliefs), which themselves
refer to state properties concerning the organization structure and to a dynamic (leads to) property
again. An example property is the following, describing that a role performs the behavior it
believes that is expected from the role:
24 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
If at time t
a role believes that
this role has as part of its behavior description that
upon input v the output action w is done,
and
v occurs as input,
then
at a next point in time this role will provide output w.
More formal details can be found in Appendix C.
6 Simulation of the Case Study: the Eleven Cities Tour example
This Section presents a case study to illustrate the usage of the organizational change model as
presented in the previous Sections. First, the organization under investigation is explained and
thereafter simulation results are presented as well as domain specific properties that have been
used to enable a simulation.
6.1 Case study description
The organization model of organizational change has been applied to the organization that is
responsible for the famous Frisian skating tour called the Eleven Cities Tour. The association is
called “De Friesche Elf Steden” in Dutch.
Although the association has fixed parts in the organization, it also has an annual dynamics in
its structure. The association has a board consisting of 3 members namely the Chairperson, the
Treasurer, and the Secretary. The Board has two responsibilities: running the association smoothly
at all times and organizing the tour. Most of the year only the board is active, but there is also a
permanent group which contains all members of the eleven cities tour society, which includes the
people within the Board as well. This off-season organization is shown in Figure 8. Once a year, at
the beginning of winter, the organization changes its structure by formation of Region groups and
the election of Region Heads for the coming winter season to enable monitoring of the ice
conditions. This change process takes place within the Eleven Cities Tour Society group where the
Modeling Centralized Organization of Organizational Change 25
Member with a shared allocation to the Chairperson in the board is in charge of the change
process. In the real organization, 21 Region groups are formed, for the case study however only
the groups for the cities of Woudsend and Sneek are assumed to be created. The Region groups
consist of more roles than the Region Head role (Monitor roles), however these roles have been
left out of the case study for the sake of clarity. The election of the Region Heads is always a
difficult part of the organizational change, as many people resist to the election of certain people
because they think these people are not suitable for the job or because they prefer another
candidate but in this case study we only consider suitability. Once the Regions have been formed
and the Region Heads have been appointed, they start their work of monitoring the ice condition
along the route. After certain conditions are met, such as a certain period of frost, another change
occurs within the organization: A group called Region Representatives is formed which consists of
representatives of all Region groups and representatives of the Board. This group discusses the
conditions along the entire trajectory of the tour. If the conditions are good, this group organizes
the Tour. The organization after formation of the Regions and Region Representatives group is
shown in Figure 9. Note that the shared allocations between the members of the Board and the
representatives of these in the Region Representatives group have been omitted to keep the Figure
clear. To ensure that indeed all roles within the Region Representatives group is according to the
Fig. 8. Off-season eleven cities tour organization
26 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
desired behavior, the Chairperson Representative monitors whether the representatives of each of
the Regions are indeed behaving according to the specification and do not fall back into their prior
behavior.
Finally, at the end of the winter, the Chair of the Meeting of Region Heads thanks all
participants and deactivates all roles in that group as well as all Region Head role instances. At this
point in time the agents are de-allocated from their roles, and the roles immediately cease to exist.
The involved agents only remain allocated to the continuous roles / roles instances in the Board
and Eleven City Tour Society group.
6.2 Simulation Results
Based on the generic properties as specified in Section 4, a domain specific simulation model for
the eleven cities tour has been created. All of the properties that underline the basis of this model
have been specified in the leadsto format as introduced in Section 3. Since this format is
executable, simulations can be performed using the leadsto software tool (Bosse et al. 2005). This
Section presents a selection of the simulation results, and gives example of the domain specific
properties that have been used for the simulation. Furthermore, several events are put into the
Fig. 9. Eleven cities tour organization after formation of the Regions and Region
Representatives groups.
Modeling Centralized Organization of Organizational Change 27
model to see how well the organization changes in case this is required. The results have been
ordered based on the different phases in organizational change distinguished by Lewin.
6.2.1 Initial organization
The initial organizational setup for the simulation is shown in Figure 10. On the left hand side of
the Figure statements are shown about the organization whereas the right hand side presents a
timeline where a black box indicates that the statement is true at that particular time point. The
Eleven Cities Tour Society group is called Change Group within the initial organization since this
group’s only function is organizational change, hence it is considered a Change Group. Note that
the Figure only presents part of the initial organization: only a selection of the intra and inter group
interactions, and only the beliefs of the Change Manager are shown. The Figure for example
shows the presence of the role Chairperson:
internal(GlobalChangeManager|ChangeGroup)|belief(exists_role(Chairperson))
Furthermore, the existence of the group Board is shown:
internal(GlobalChangeManager|ChangeGroup)|belief(exists_group(Board))
The role Chairperson is specified to be part of the Board group:
internal(GlobalChangeManager|ChangeGroup)|belief(role_belongs_to_group(Chairperson, Board))
internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(Chairperson))internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(GlobalChangeManager))
internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(MemberFive))internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(MemberFour))internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(MemberOne))
internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(MemberThree))internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(MemberTwo))
internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(Secretary))internal((GlobalChangeManager|ChangeGroup))|belief(exists_role(Treasurer))
internal((GlobalChangeManager|ChangeGroup))|belief(exists_group(Board))internal((GlobalChangeManager|ChangeGroup))|belief(exists_group(ChangeGroup))
internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(Chairperson, Board))internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(Secretary, Board))internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(Treasurer, Board))
internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(GlobalChangeManager, ChangeGroup))internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(MemberFive, ChangeGroup))internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(MemberFour, ChangeGroup))internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(MemberOne, ChangeGroup))
internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(MemberThree, ChangeGroup))internal((GlobalChangeManager|ChangeGroup))|belief(role_belongs_to_group(MemberTwo, ChangeGroup))
internal((GlobalChangeManager|ChangeGroup))|belief(intra_group_connection(Chairperson, Secretary, Board, t1))internal((GlobalChangeManager|ChangeGroup))|belief(intra_group_connection(Secretary, Chairperson, Board, t2))
internal((GlobalChangeManager|ChangeGroup))|belief(inter_group_connection(Chairperson, Board, GlobalChangeManager, ChangeGroup, gi1))internal((GlobalChangeManager|ChangeGroup))|belief(inter_group_connection(Secretary, Board, MemberFour, ChangeGroup, gi2))
time 0 1 2 3 4 5
Fig. 10. Initial setup of the organization for the simulation.
28 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
Intra group interaction is part of the Board group as well, the Secretary can for example
communicate with the Chairperson:
internal(GlobalChangeManager|ChangeGroup)|belief(intra_group_connection(Secretary, Chairperson,
Board, t1))
And finally, inter group connections are part of the beliefs of the Change Manager. The inter group
connection shown is the one between the Chairperson in the Board and the Change Manager
within the Change Group:
internal(GlobalChangeManager|ChangeGroup)|belief(inter_group_connection(Chairperson, Board,
GlobalChangeManager, ChangeGroup, gi1))
This inter group connection is based on shared allocation, which means that the agent playing the
role of Chairperson within the Board also plays the role of Change Manager within the Change
Group. Within the Figure, the Change Group consists of five Member roles and one Change
Manager. The additional Members in the Change Group are played by agents that are not yet part
of the organization, but can be used by the Change Manager for the fulfillment of new roles to be
played.
6.2.2 Unfreezing phase for region formation
After the initial setup of the organization, an event is put into the simulation which requires an
organizational change, namely the onset of winter meaning that it is time to form the regions
within the organization. The first phase within this change process is unfreezing. The occurrences
during this phase are shown in Figure 11. The event requiring change is the Chairperson within the
board observing that it is time to form the regions:
input((Chairperson|Board))|time_to_form_regionsoutput((GlobalChangeManager|ChangeGroup))|inform(change_group_active)
output((GlobalChangeManager|ChangeGroup))|inform(organizational_model(region_structure))input((MemberOne|ChangeGroup))|inform(organizational_model(region_structure))input((MemberTwo|ChangeGroup))|inform(organizational_model(region_structure))
input((GlobalChangeManager|ChangeGroup))|accept(organizational_model(region_structure), MemberOne, ChangeGroup)input((GlobalChangeManager|ChangeGroup))|accept(organizational_model(region_structure), MemberTwo, ChangeGroup)
output((GlobalChangeManager|ChangeGroup))|request_candidates_for_regionsinput((GlobalChangeManager|ChangeGroup))|proposal(MemberOne, RegionHeadSneek, RegionSneek)
input((GlobalChangeManager|ChangeGroup))|proposal(MemberTwo, RegionHeadWoudsend, RegionWoudsend)output((GlobalChangeManager|ChangeGroup))|inform(shared_allocation, MemberOne, RegionHeadSneek, RegionSneek)
output((GlobalChangeManager|ChangeGroup))|inform(shared_allocation, MemberTwo, RegionHeadWoudsend, RegionWoudsend)output((MemberTwo|ChangeGroup))|accept(shared_allocation, MemberTwo, RegionHeadWoudsend, RegionWoudsend)
input((GlobalChangeManager|ChangeGroup))|oppose(inform(shared_allocation, MemberOne, RegionHeadSneek, RegionSneek), MemberTwo, not_suitable_candidate)output((GlobalChangeManager|ChangeGroup))|additional_argument(inform(shared_allocation, MemberOne, RegionHeadSneek, RegionSneek), only_candidate)
output((MemberTwo|ChangeGroup))|accept(shared_allocation, MemberOne, RegionHeadSneek, RegionSneek)time 0 5 10 15 20 25
Fig. 11. First unfreezing phase during simulation
Modeling Centralized Organization of Organizational Change 29
input(Chairperson|Board)|time_to_form_regions
An inter-group interaction property in the form of a leadsto rule now fires which specifies that if
the Chairperson within the Board observes it is time to form the regions, the Change Manager
activates the Change Group and announces the organizational model for the region structure:
GIP_specific(Chairperson, Board, ChangeManager, ChangeGroup): Form Regions when winter
if at time t the Chairperson within the Board observes that it is time to form the Region groups
then at time t + 1 the Global Change Manager within the Change Group informs the Members within the Change Group
that the group is now active
and at time t + 1 the Global Change Manager within the Change Group announces the new organizational model
regarding the Regions.
The results of this rule show in the trace by the following elements:
output(GlobalChangeManager|ChangeGroup)|inform(change_group_active)
output(GlobalChangeManager|ChangeGroup)|inform(organizational_model(region_structure))
Only a reference i.e. the statement region_structure, to the whole specification of this organizational
structure is presented in the Figure for the sake of clarity. None of the members oppose the change
as all are skating fanatics that long for a tour and all are convinced that winter has started. For
them the onset of winter naturally means the formation of regions, so all communicate the
acceptance of the organizational model:
input(GlobalChangeManager|ChangeGroup)|accept(organizational_model(region_structure), MemberOne,
ChangeGroup)
The unfreezing for this particular organizational structure is therefore accomplished, following
RP2 as described in Section 4. Another element of the change is to allocate the appropriate agents
to the specific roles within the new organizational model: appoint the Region Heads. For this a
more complicated unfreezing phase is performed. First, the Change Manager within the Change
Group requests candidates for the newly formed roles:
output(GlobalChangeManager|ChangeGroup)|request_candidates_for_regions
The Members within the ChangeGroup receive the request and propose candidates for the
positions, based upon their availability during the winter:
input(GlobalChangeManager|ChangeGroup)|proposal(MemberOne, RegionHeadSneek, RegionSneek) input(GlobalChangeManager|ChangeGroup)|proposal(MemberTwo, RegionHeadWoudsend,
RegionWoudsend)
30 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
After receiving the proposals, the Change Manager decides upon an optimal allocation. Since there
are two roles that need to be fulfilled and there is one proposal per role, these allocations are
chosen and communicated:
output(GlobalChangeManager|ChangeGroup)|inform(shared_allocation, MemberOne, RegionHeadSneek,
RegionSneek) output(GlobalChangeManager|ChangeGroup)|inform(shared_allocation, MemberTwo,
RegionHeadWoudsend, RegionWoudsend)
Member Two is however not convinced about the suitability of Member One for the role of
Region Head Sneek and opposes to the organizational model following a domain specific
instantiation of RP3 in Section 4:
RP3_specific(MemberTwo): Oppose to Change
if at time t Member Two is informed about a change to an organizational model M in which a Member M1
has a shared allocation to a Role R1
and Member Two observes M1 is unsuitable for the Role R1 at time t
then at time t + 1 Member Two opposes to the change to the organizational model stating that M1 is not suitable for R1
The result of this rule is shown in the trace by the following statement:
input(GlobalChangeManager|ChangeGroup)|oppose(inform(shared_allocation, MemberOne,
RegionHeadSneek, RegionSneek), MemberTwo, not_suitable_candidate)
As a result a domain specific instantiation of RP4 fires which is specified below.
RP4_specific(ChangeManager): Convince Member
if at time t a Member M1 communicates opposing to the change to organizational model M to the Change
Manager because candidate M2 is considered not suitable for the allocation to role R1
and the Change Manager observed M2 is the only candidate for the role R1 at time t
then at time t + 1 the ChangeManager communicates that M2 is the only candidate for role R1.
In the trace, the communication can be seen in the following format:
output(GlobalChangeManager|ChangeGroup)|additional_argument(inform(shared_allocation, MemberOne,
RegionHeadSneek, RegionSneek), MemberTwo, only_candidate)
Finally, a rule RP5 is specified for this domain as well, as shown below. Since the successful
organization of an eleven cities tour is most important for the Members and all roles being
allocated is essential for such a successful organization, they seize to oppose to an allocation in
case they are informed about the existence of only one candidate.
Modeling Centralized Organization of Organizational Change 31
RP5_specific(Member Two): Member Convinced
if at time t Member Two opposes to the change to the organizational model M regarding the allocation of
Member M1 to Role R1
and at time t2 later than t Member Two receives the argument that Member M1 is the only candidate for role R1
then at time t2 + 1 Member Two will inform the Change Manager upon its acceptance of the change to the
organizational model M
In the trace the Member indeed outputs the belief upon the shared allocation:
output(MemberTwo|ChangeGroup)|accept(shared_allocation, MemberOne, RegionHeadSneek,
RegionSneek)
Since all Members have now communicated their acceptance of the new organizational model, the
unfreezing phase is performed successfully.
6.2.3 Movement and Refreezing of the region formation
The movement and refreezing phase for the case study are much shorter than the unfreezing phase,
as the new organizational model is already accepted by all Members of the organization. The two
phases are shown in Figure 12. Trigger for the ChangeManager to start the movement phase is
when an acceptance on all parts of the organizational model M has been communicated to the
Change Manager, as specified before in RP6. The movement phase starts with the communication
of the region structure being active:
output(GlobalChangeManager|ChangeGroup)|inform(active(organizational_model(region_structure)))
The phase ends after all participants of the change have confirmed that the organizational model
will be active, which they instantly do as they are already unfrozen:
input(GlobalChangeManager|ChangeGroup)|accept(active(organizational_model(region_structure)),
MemberOne, ChangeGroup)
input(GlobalChangeManager|ChangeGroup)|accept(active(organizational_model(region_structure)),
MemberTwo, ChangeGroup)
Finally, the refreezing phase ends after the duration set by the Change Manager. In this particular
refreezing phase, all roles immediately behave correctly after the change (according to RP7 in
output((GlobalChangeManager|ChangeGroup))|inform(active(organizational_model(region_structure)))input((GlobalChangeManager|ChangeGroup))|accept(active(organizational_model(region_structure)), MemberOne, ChangeGroup)input((GlobalChangeManager|ChangeGroup))|accept(active(organizational_model(region_structure)), MemberTwo, ChangeGroup)
output((GlobalChangeManager|ChangeGroup))|inform(change_group_inactive)time 0 5 10 15 20 25 30
Fig. 12. First movement and Refreezing
32 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
Section 4.2.3) which is not shown in the trace for the sake of brevity. Eventually, the Change
Group is deactivated:
output(GlobalChangeManager|ChangeGroup)|inform(change_group_inactive)
6.2.4 Unfreezing phase for regions representatives group
The second unfreezing phase which is required to form the Region Representatives group is shown
very briefly in Figure 13. As a start of the unfreezing phase the following events are put into the
simulation:
input(Chairperson|Board)|one_week_frost_period_just_passed
input(Chairperson|Board)|before_that_week_no_frost
input(Chairperson|Board)|no_tour_held_this_winter
As a result, the following inter group interaction property fires:
GIP_specific(Chairperson, Board, ChangeManager, ChangeGroup): Form Region Representatives group after frost
period
if at time t the Chairperson within the Board observes a period of one week of frost
and at time t – (1 week) the Chairperson within the observed that there was no frost
and at no time point this year the Chairperson within the Board observed that a tour has been held
then at time t + (1 day) the Change Manager within the Change Group informs the Members within the Change
Group that the group is now active
and at time t + (1 day) the Change Manager within the Change Group announces the new organizational model
regarding the Regions Representatives group.
The resulting communication of the Change Manager is shown in the trace: A communication of
the change group being active again, and communication of the new organizational model:
output(GlobalChangeManager|ChangeGroup)|inform(change_group_active)
output(GlobalChangeManager|ChangeGroup)|inform(organizational_model(region_coordination_structure))
input((Chairperson|Board))|before_that_week_no_frostinput((Chairperson|Board))|no_tour_held_this_winter
input((Chairperson|Board))|one_week_frost_period_just_passedoutput((GlobalChangeManager|ChangeGroup))|inform(Change_group_active)
output((GlobalChangeManager|ChangeGroup))|inform(organizational_model(region_representatives_structure))input((GlobalChangeManager|ChangeGroup))|accept(organizational_model(region_representatives_structure), MemberOne, ChangeGroup)input((GlobalChangeManager|ChangeGroup))|accept(organizational_model(region_representatives_structure), MemberTwo, ChangeGroup)
time 0 20 40 60 80
Fig. 13. Second unfreezing phase
Modeling Centralized Organization of Organizational Change 33
All Members accept the new structure, as they are very eager just thinking about a possible eleven
cities tour, the event of the year, and are therefore immediately unfrozen, communicating their
acceptance to the Change Manager. Therefore, the unfreezing is performed using RP2. Resistance
can however easily be incorporated using properties such as presented in Section 6.2.2. The
unfreezing process has now ended successfully.
6.2.5 Movement and Refreezing of the region representatives group
output((GlobalChangeManager|ChangeGroup))|inform(active(organizational_model(region_representatives_structure)))
input((GlobalChangeManager|ChangeGroup))|accept(active(organizational_model(region_representatives_structure)), MemberOne, ChangeGroup) input((GlobalChangeManager|ChangeGroup))|accept(active(organizational_model(region_representatives_structure)), MemberTwo, ChangeGroup)
internal((RegionHeadWoudsend|RegionWoudsend))|belief(has_expression(gip1, leads_to((input((RegionHeadWoudsend|RegionWoudsend))|report(RegionWoudsend, good)), (output((RegionRepresentativeWoudsend|RegionRepresentatives))|report(RegionWoudsend, good)))))
internal((RegionHeadWoudsend|RegionWoudsend))|belief(exists_role(RegionRepresentativeWoudsend)) internal((RegionHeadWoudsend|RegionWoudsend))|belief(exists_group(RegionRepresentatives))
internal((RegionHeadWoudsend|RegionWoudsend))|belief(role_belongs_to_group(RegionRepresentativeWoudsend, RegionRepresentatives)) internal((RegionHeadWoudsend|RegionWoudsend))|belief(group_interaction_property(gip1, RegionHeadWoudsend, RegionWoudsend, RegionRepresentativeWoudsend, RegionRepresentatives))
internal((RegionHeadWoudsend|RegionWoudsend))|belief(inter_group_connection(RegionHeadWoudsend, RegionWoudsend, RegionRepresentativeWoudsend, RegionRepresentatives, gi24)) output((GlobalChangeManager|ChangeGroup))|inform(Change_group_inactive)
input((RegionHeadWoudsend|RegionWoudsend))|report(RegionWoudsend, good) input((RegionHeadSneek|RegionSneek))|report(RegionSneek, good)
output((RegionRepresentativeSneek|RegionRepresentatives))|report(RegionSneek, good) output((CharipersonRepresentative|RegionRepresentatives))|remind(gip1)
output((RegionRepresentativeWoudsend|RegionRepresentatives))|report(RegionWoudsend, good) output((Chairperson|Board))|let_the_tour_be_held_on_date
Time
0 20 40 60 80
Fig. 14. Second Movement and refreezing
After unfreezing the organization, the Change Manager communicates that the new organization
with the new Region Representatives structure is now active, which is shown in the partial trace in
Figure 14:
output(GlobalChangeManager|ChangeGroup)|inform(active(
organizational_model(region_coordination_structure))
As a result, the organizational model becomes active in the actual organization, not only in the
internal beliefs of the Members of the Change Group. Within the simulation there is a mapping
between the name of general organizational structures (e.g. region coordination structure) and the
actual changes on a lower level. For the Region Head Woudsend for example, the internal belief
that a new role Region Representative Woudsend exists is added:
internal((RegionHeadWoudsend|RegionWoudsend))|belief(exists_role(RegionRepresentativeWoudsend))
Furthermore, the group Region Representatives is added to the internal beliefs:
internal((RegionHeadWoudsend|RegionWoudsend))|belief(exists_group(RegionRepresentatives))
The role RegionRepresentativeWoudsend belongs to the group RegionRepresentatives:
internal((RegionHeadWoudsend|RegionWoudsend))|belief(role_belongs_to_group(
RegionRepresentativeWoudsend, RegionRepresentatives))
34 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
A belief on an inter-group connection is added between the Region Head Woudsend within the
Region Woudsend and the Region Representative Woudsend within the Region Representatives:
internal((RegionHeadWoudsend|RegionWoudsend))|belief(inter_group_connection(RegionHeadWoudsend,
RegionWoudsend, RegionRepresentativeWoudsend, RegionRepresentatives, gi24))
Besides the structure itself, the new roles also require new behavior. When first starting to perform
a new role, the new behavior associated with the role is far from automated, and requires internal
beliefs on the desired behavior. Such elements are shown in the trace of Figure 9 as well. First of
all, there is an internal belief about the existence of a group interaction property gip1:
internal((RegionHeadWoudsend|RegionWoudsend))|belief(group_interaction_property(gip1,
RegionHeadWoudsend, RegionWoudsend, RegionRepresentativeWoudsend, RegionRepresentatives))
The specification of the behavior required by such a property is done using a TTL expression,
more particular in leadsto format:
internal((RegionHeadWoudsend|RegionWoudsend))|belief(has_expression(gip1,
leads_to((input((RegionHeadWoudsend|RegionWoudsend))|report(RegionWoudsend, good)),
output((RegionRepresentativeWoudsend|RegionRepresentatives))|report(RegionWoudsend, good),
efgh(0,0,1,1))))
This specifies that if the Region Head Woudsend receives a report that the ice is good, then this
will be communicated by the Region Representative Woudsend in the Region Representatives
group as well, with an efgh value of (0,0,1,1). Around time point 65 the antecedent of this rule
becomes true, however, the consequent is not true after 1 time point within the Region
Representatives group. As a result, the Chairperson Representative within the Region
Representatives group reminds the role of the desired behavior gip1 (according to RP9 in Section
4.2.3):
output((CharipersonRepresentative|RegionRepresentatives))|remind(gip1)
After having received this reminder, the Region Representative Woudsend does behave according
to gip1 and outputs the consequent:
output((RegionRepresentativeWoudsend|RegionRepresentatives))|report(RegionWoudsend, good)
This refreezing therefore takes the form of GP6 (Section 4.2.3) and the properties below it in the
property tree. In exceptional years, all Region Representatives report that the ice is good, and the
Chairperson within the board announces the date the tour will take place:
output((Chairperson|Board))|let_the_tour_be_held_on_date
Modeling Centralized Organization of Organizational Change 35
7 Verification of the Case Study Simulation
As for verification of the organization process of the Eleven Cities Tour is concerned, a
distinction is made between two types of verification. Firstly, guarantees are given that concern the
tour itself (so-called content properties). For example, it the circumstances permit so (if the ice is
thick enough over the whole trajectory) then a tour should be organized as soon as possible.
Secondly, guarantees on the organization of organizational change for setting up the tour are
verified (so-called organizational change properties). This Section presents both verification types.
Logical relationships between properties, as depicted in the tree of Section 4, can be very
useful in analyzing the dynamic properties of an organization. For example, if for a given trace of
the system some global property OP is not satisfied, then by a refutation process it can be
concluded that either one of the group properties, or one of the group interaction properties in the
tree does not hold. If, after checking these properties, it turns out that a group property does not
hold, then either one of the role properties or the intra group interaction properties is not satisfied.
By this refutation analysis it follows that if OP does not hold for a given trace, then, via the
intermediate properties, the cause of this malfunctioning can be found in the set of leaves of the
tree of Section 4.
In order to determine which one of the properties encountered in this refutation process
actually is refuted, some mechanism is needed to check if a certain property holds for a given
trace. To this end, the simulation software described in Section 6 automatically produces log files
containing the traces. In addition, software has been developed that is able to read in these log files
together with a set of dynamic properties (in TTL format), and to perform the checking process.
Traces are thus analyzed with an automated logic-based checker. This checker takes as input a
property of interest about the trace and logically validates whether the property holds in the given
trace. If the property holds in the trace, the checker outputs success otherwise it outputs failure.
But the software determines not only whether the properties hold for the trace or not, but in case of
failure, it also pinpoints which parts of the trace violate the properties. The results of different
checks that have been performed are described below.
36 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
7.1 Content Properties
The overall goal of the Eleven Cities Tour organization is to arrange for a tour to be organized
when possible, i.e., when the ice along the tour is thick enough to ensure a safe passage. This
following property expresses this goal: the tour has to be organized whenever possible, ensuring a
safe passage over the ice for all skaters.
OP3: Organize tour in case of good conditions
if the ice conditions in all regions are good
then it is announced that the tour will be held
This property has been checked against the simulation trace that was presented in the previous
Section and is indeed satisfied within that trace. Other content properties to consider in this context
are, for example, the organization daily decides on the possibility and date (if appropriate) of a
tour: ‘ it giet oan’ (in Frisian language a go decision) decisions, and in wintertime, the organization
daily monitors the weather. However, only OP3 is addressed in this paper.
7.2 Organizational Change Properties
The properties as presented in the previous Section depend on some organizational structure to
ensure the fulfillment of each property and all of them combined. For this purpose, the aim of this
paper is exactly this: a way to specify and model such an organization itself has been presented, as
well as the actual process of setting up the organization. As such, this organization can support the
organizational properties as presented above.
For the purpose of verifying the organizational change in the Eleven Cities Tour simulation,
automatic checking of the high-level properties presented in Section 4.2 has been performed on the
generated trace. The results are shown in Table 6. In the simulation trace, there are 2 change
moments: the formation of the regions structure and the formation of the region representatives
group. For the last changes, there is no resistance to the organization change, while there is for the
first one. The automated checker has verified that all properties specifying a successful phase are
indeed satisfied, hence, both changes have passed a successful unfreezing, movement, and
refreezing phase. There is however a difference in how this success was accomplished. In the first
change, property GP3 was satisfied, specifying that there was resistance to the change which was
Modeling Centralized Organization of Organizational Change 37
taken away. In the second change however, the change went without resistance; property GP3 was
not satisfied in that change. In the refreezing phase of the first change, property GP6 was not
satisfied as no improper behavior was encountered. In the second change however, improper
behavior did show, after which the behavior was corrected, satisfying property GP6. The following
setting were used for checking: For the unfreezing phase e was set to 1.0. Regarding the refreezing
phase both e1 and e2 have been set to 1.0, for p1 a value of 10 was used, and finally, p2 was set to
20.
Table 6. Checked Properties (Yes = satisfied, No = not satisfied)
Stage Property Change 1 Change 2
Unfreezing GP1 Yes Yes
GP3 Yes No
Moving OP1 Yes Yes
Refreezing OP2 Yes Yes
GP6 No Yes
8 Conclusions
Organizations often have to survive in a dynamic world. To enable organizations in practice to
adapt to the dynamics of the world, certain facilities, structures and capabilities are needed that
support organizational change. This paper shows how the organization of organizational change
processes can be modeled within a formal organization modeling approach. A generic organization
model for organizational change was presented and formally verified for a case study concerning
the organization of a major event in the Netherlands: the eleven cities tour. The formal verification
sets it apart from existing work on organization modeling, e.g., (Fox and Gruninger 1998; Steen
Lankhorst and Wetering 2002). Previous work of the authors on organizational change (Jonker
Schut and Treur 2003) considered change as an instantaneous event instead of a process of change
as is done in this paper. Additionally, previous work did not include the distinction between formal
languages for expressing the change process. The change model in this paper takes into account
different phases in a change process (unfreezing, movement and refreezing) considered in (Lewin
38 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
1951), which is still considered valid in current organizational change literature, see e.g. (Robbins
1998; Orlikowski and Hofman 1997). In (Orlikowski and Hofman 1997) a distinction is made
between anticipated change (for which the model of Lewin is said to be suitable), emergent change
and opportunity-based change. In this paper only anticipated change is being modeled and
therefore the other two types of change are not addressed. In change processes the internal
(mental) states of those involved in the organization are important. Therefore, also internal states
of individuals have to be part of a model for organizational change. In particular, beliefs and their
changes have been incorporated in the model. In addition, an internal model for (reflective)
reasoning about expected role behavior was included. Hence, a model was created that combines
organization aspects and cognitive aspects.
Acknowledgements
This research has been performed as part of two projects: CIM, for Cybernetic Incident
Management, and DEAL, for Distributed Engine for Advanced Logistics. Both projects are
funded by the Dutch Ministry of Economic Affairs. Furthermore, the authors would like to
thank the anonymous reviewers for their useful comments that helped to improve the paper.
References
Abbink, H., Dijk, R. van, Dobos, T., Hoogendoorn, M., Jonker, C.M., Konur, S., Maanen, P.P.
van, Popova, V., Sharpanskykh, A., Tooren, P. van, Treur, J., Valk, J., Xu, L., Yolum, P., (2004).
Automated Support for Adaptive Incident Management. In: Walle, B. van de, and Carle, B.
(eds.), Proc. of the First International Workshop on Information Systems for Crisis Response
and Management, ISCRAM'04, pp. 69-74..
Ackerman, L.S., Development, transition, or transformation: the question of change in
organization, OD Practitioner, 18(4) , 1986, pp. 1-9.
Alvesson, M., Cultural Perspectives on Organizations, Cambridge University Press, New York,
1993.
Bacharach, S.B. and Lawler, E.J., Power and politics in organizations, Jossey-Bass, San Francisco,
1980.
Modeling Centralized Organization of Organizational Change 39
Bashein, M.L., Marcus, M.L., and Riley, P., Business Process Reengineering: preconditions for
success and failure, Informations Systems Management 9, 1994, pp. 24-31.
Boonstra, J.J. (editior), Dynamics of Organizational Change and Learning, Wiley, 2004.
Bosse, T., Jonker, C.M., Mey, L. van der, and Treur, J., LEADSTO: a Language and Environment
for Analysis of Dynamics by SimulaTiOn. In: Eymann, T., et al. (eds.), Proc. of the Third
German Conference on Multi-Agent System Technologies, MATES'05. Lecture Notes in
Artificial Intelligence, vol. 3550. Springer Verlag, 2005, pp. 165-178.
Bradshaw, P., and Boonstra, J.J., Power Dynamics in Organizational Change, In: Boonstra, J.J.
(editor), Dynamics of Organizational Change and Learning, Wiley, 2004, pp. 279-299.
Cummings, T.G., and Worley, C.G., Organization Development and Change, South Western
College Publishing, 2001.
Cummings, T.G., Organization Development and Change, In: Boonstra, J.J. (editor), Dynamics of
Organizational Change and Learning, Wiley, 2004, pp. 25-42.
Dahl, R.A., The concept of power, Behavioral Science, 2, 1975, pp. 201-215.
Emerson, R.M., Power dependence relations, American Sociological Review, 27, 1962, pp. 31-41.
Ferber, J. and Gutknecht, O. (1998), “A meta-model for the analysis and design of organizations in
multi-agent systems,“ Proceedings of the Third International Conference on Multi-Agent
Systems (ICMAS’98), IEEE Computer Society Press, pp. 128-135.
Ferber, J., Gutknecht, O., Jonker, C.M., Müller, J.P., and Treur, J. (2001), ”Organization Models
and Behavioural Requirements Specification for Multi-Agent Systems,“ in Y. Demazeau, F.
Garijo (Eds.), Multi-Agent System Organizations. Proceedings of MAAMAW'01.
Fox, M.S., and Gruninger, M. (1998), “Enterprise Modelling,” . AI Magazine, 19(3), AAAI Press,
pp. 109-121.
Glaser, N., and Morignot, P. (1997), “The Reorganization of Societies of Autonomous Agents,” in
Boman, M. Velde, W. van de (eds.), Multi-Agent Rationality, 8th European Workshop on
Modelling Autonomous Agents in a Multi-Agent World, Lecture Notes in Computer Science,
vol. 1237, Springer, pp. 98-111.
Hall, G., Rosenthal, T., and Wade, J. How to make reengineering really work, Harvard Business
Review, 71(6), 1993, pp. 119-131.
Hoogendoorn, M., Jonker, C. M., Konur, S., Maanen, P.P. van, Popova, V., Sharpanskykh, A.,
Treur, J., Xu, L., Yolum, P., (2004). Formal Analysis of Empirical Traces in Incident
Management, In: Macintosh, A., Ellis, R., and Allen, T. (eds.), Applications and Innovations in
40 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
Intelligent Systems XII, Proceedings of AI-2004, the 24th SGAI International Conference on
Innovative Techniques and Applications of Artificial Intelligence. Springer Verlag, 2004, pp.
237-250.
Hoogendoorn, M., Jonker, C.M., Popova, V., Sharpaskykh, A., Xu, L., Formal Modelling and
Comparing of Disaster Plans. (2005). In: Carle, B., and Walle, B. van de, (eds.), Proceedings of
the Second International Conference on Information Systems for Crisis Response and
Management ISCRAM '05, pp. 97-107.
Hosking, D.M., Social construction as process: some new possibilities for research and
development, Concepts and Transformation, 4(2), 1999, pp. 117-132.
Huczynski, A., and Buchanan, D. (2001), Organizational Behaviour, Prentice Hall.
Jaffee, D. (2001), Organization Theory – Tension and Change, McGraw-Hill Companies.
Jonker, C.M., and Treur, J. (2002), “Compositional Verification of Multi-Agent Systems: a Formal
Analysis of Pro-activeness and Reactiveness,” International Journal of Cooperative Information
Systems, 11, pp. 51-92.
Jonker, C.M., and Treur, J. (2003), “Relating Structure and Dynamics in an Organization Model,”
In J.S. Sichman, F. Bousquet, and P. Davidson (eds.), Multi-Agent-Based Simulation II,
Proceedings of the Third International Workshop on Multi-Agent Based Simulation, MABS'02.
Lecture Notes in AI, vol. 2581, Springer Verlag, pp. 50-69.
Jonker, C.M., Schut, M.C., and Treur, J. (2003), “Modelling the Dynamics of Organizational
Change,” In Klusch, M., Omicini, A., and Ossowski, S., and Laamanen, H., (eds.), Cooperative
Information Agents VII, Proceedings of the Seventh International Workshop on Cooperative
Information Agents, CIA 2003, Lecture Notes in AI, vol. 2782, Springer Verlag, pp. 336-344.
Kotter, J.P., Leading Change, Harvard Business School Press, Boston, 1999.
Lewin, K. (1951), Field Theory in Social Science, Harper & Row, New York.
Lippit, R., Watson, J., Westley, B., The Dynamics of Planned Change, Harcourt, New York, 1958.
Orlikowski, W. and Hofman, D. (1997), “An Improvisational Model of Change Management: The
Case of Groupware Technologies,” Sloan Management Review vol. 38 (2), pp 11-22.
Robbins, S.P. (1998), Organizational Behaviour, Prentice Hall, New Jersey.
Schein, E.H., On dialogue, culture, and organizational learning, Organizational Dynamics 22(2),
1993, pp. 40-51.
Modeling Centralized Organization of Organizational Change 41
Steen, M.W.A., Lankhorst, M.M. and Wetering, R.G. van de (2002), “Modelling Networked
Enterprises,” Proceedings of the 6th International Enterprise Distributed Object Computing
Conference (EDOC), IEEE Computer Society, pp. 109-119.
Wrong, D.H., Some problems in defining social power, American Journal of Sociology, 73, 1968,
pp. 673-681.
Author Bios
Mark Hoogendoorn is a PhD student at the Vrije Universiteit Amsterdam, Department of
Artificial Intelligence. He obtained his Masters degree in Computer Science in 2003 at the
same university, graduating on a project related to multi-agent negotiation as part of the
MAGNET research group at the University of Minnesota. In his PhD research he focuses on
organizational change within multi-agent systems, applying his research in various domains,
including incident management, logistics, and the naval domain.
Catholijn M. Jonker is a full professor in Artificial Intelligence and Cognitive Science at
the Nijmegen Institute for Cognition and Information of the Radboud Universiteit Nijmegen
in the Netherlands. She studied computer science at Utrecht University. She completed her
PhD on the topic of Negations and Constraints in Logic Programming at the same university.
After completing a post-doc position on the same topic at the Universität Bern, she became
an assistant professor at the Vrije Universiteit Amsterdam and switched her research topic to
agent technology. During the time at the Vrije Universiteit her interest in cognitive science
increased, which she combined with her work on modeling multi-agent systems and
organizations and her work on the analysis and modeling of the dynamics of behavior of
complex systems. These research topics now contribute to the research program of the
Cognitive Artificial Intelligence division of the Nijmegen Institute for Cognition and
Information.
Martijn C. Schut is Assistant Professor at the Department of Artificial Intelligence, Vrije
Universiteit, Amsterdam, The Netherlands. He received a MSc from the Vrije Universiteit
42 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
(NL) and a PhD from the University of Liverpool (UK). His research interests concern the
emergence of organizational dynamics in distributed multi-agent systems.
Jan Treur received his Ph.D. in Mathematics and Logic in 1976 from Utrecht University.
Since 1986 he works in Artificial Intelligence, from 1990 as a full professor and head of the
Department of Artificial Intelligence at the Vrije Universiteit Amsterdam. In the 1990s he
headed a research program on component-based design of knowledge-based and agent
systems. In the last five years the research program focused on modeling dynamics of agent
systems in practical application areas, and related to other disciplines such as Biology,
Cognitive Science, Organization Theory, and Philosophy of Mind.
Modeling Centralized Organization of Organizational Change 43
Appendix A Temporal Trace Language TTL: More Formal Details
A state ontology is a specification (in order-sorted logic) of a vocabulary. A state for ontology
Ont is an assignment of truth-values {true, false} to the set At(Ont) of ground atoms expressed in
terms of Ont. The set of all possible states for state ontology Ont is denoted by STATES(Ont). The
set of state properties STATPROP(Ont) for state ontology Ont is the set of all propositions over
ground atoms from At(Ont). A fixed time frame T is assumed which is linearly ordered. A trace or
trajectory γ over a state ontology Ont and time frame T is a mapping γ : T → STATES(Ont), i.e., a
sequence of states γt (t ∈ T) in STATES(Ont). The set of all traces over state ontology Ont is denoted
by TRACES(Ont). Depending on the application, the time frame T may be dense (e.g., the real
numbers), or discrete (e.g., the set of integers or natural numbers or a finite initial segment of the
natural numbers), or any other form, as long as it has a linear ordering. The set of dynamic
properties DYNPROP(Ont) is the set of temporal statements that can be formulated with respect to
traces based on the state ontology Ont in the following manner.
Given a trace γ over state ontology Ont, the input state of some role r within a group g at time point
t is denoted by
state(γ, t, input(r|g))
analogously
state(γ, t, output(r|g))
state(γ, t, internal(r|g))
denote the output state and internal state.
These states can be related to state properties via the formally defined satisfaction
relation |=, comparable to the Holds-predicate in the Situation Calculus: state(γ, t, output(r|g)) |= p
denotes that state property p holds in trace γ at time t in the output state of role r within group g.
Based on these statements, dynamic properties can be formulated in a formal manner in a sorted
first-order predicate logic with sorts TIME or T for time points, Traces for traces and F for state
formulae, using quantifiers over time and the usual first-order logical connectives such as ¬, ∧, ∨,
�, ∀, ∃. In trace descriptions, notations such as state(γ, t, output(r|g))|= p are shortened to output(r|g)|p.
44 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
Appendix B Constraints on the Language Elements
Role dynamic properties
If has_expression(p:DYNPROP, d:DYNPROPEXP)
and role_property(p, r:ROLE, g:GROUP)
then element_of(d, DYNPROPEXP(r|g, ONT(r|g)))
The group is also part of the definition of the ontology since roles in different groups can have the
same name and might have a different ontology.
Role properties can be divided into different types which in turn can be defined more restricted
than the general definition. An example of such a refinement is an executable role dynamic
property. This special type is defined as follows:
if has_expression(p:DYNPROP, d:DYNPROPEXP)
and role_property(p, r:ROLE, g:GROUP)
then element_of(d, DYNPROPEXP((r|g), role_input_ontologies(r|g) ∪ role_ouput_ontologies(r|g)))
Transfer dynamic properties
If has_expression(p:DYNPROP, d:DYNPROPEXP)
and transfer_property(p, r1:ROLE, r2:ROLE, g:GROUP)
then element_of(d, DYNPROPEXP({r1|g, r2|g}, role_output_ontologies(r1|g) ∪
role_input_ontologies(r2|g)))
Group dynamic properties
If has_expression(p:DYNPROP, d:DYNPROPEXP)
and group_property(p, g:GROUP)
then element_of(d, DYNPROPEXP(g, ONT(g)))
Modeling Centralized Organization of Organizational Change 45
Intergroup interaction dynamic properties
If has_expression(p:DYNPROP, d:DYNPROPEXP)
and group_interaction_property(p, r1:ROLE, g1:GROUP, r2:ROLE, g2:GROUP)
then element_of(d, DYNPROPEXP({r1|g1, r2|g2}, role_input_ontologies(r1|g1)
∪ role_output_ontologies(r2|g2)))
46 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
Appendix C Changing Organizations Formalized in meta TTL
Here is the unchanged formal part from the previous Section 6 and a formal example from
previous Section 5.
C.1 Sorts and Subsorts in meta TTL
Table C.1. Sorts in meta TTL
Sort Description
TRACE for traces
STATE for states within a trace.
T time frame.
STATOMS expressions for state atoms.
CONSTATOMS expressions for conjunctions of state atoms.
STATPROPEXP expressions for state properties.
The sorts that are included in meta TTL are shown in Table C.1. The subsort relation
STATOMS ⊆ CONSTATOMS holds.
The function
and: CONSTATOMS x CONSTATOMS → CONSTATOMS
is used to build conjunctions of state atoms; it is also written as ∧ in infix notation
Furthermore, the relation <: T x T for time ordering is used , and the function
state: TRACE x T x PART → STATE
that indicates the state of part of the considered system within a trace at some point in time.
For the changing organization it is needed to use names and expressions for dynamic properties
within other formulae. Therefore two sorts
DYNPROP names for dynamic properties
DYNPROPEXP expressions for dynamic properties
have been introduced in the Appendix A.
Moreover,
holds: STATE x STATPROPEXP → DYNPROPEXP
Modeling Centralized Organization of Organizational Change 47
indicates the dynamic property that a state property expression is true in a state; this predicate
holds is often written as |= in infix notation.
C.2 Example formalization in change language
By means of an example the use of the functions combined with the language is shown below.
to_be_performed(delete(exists_role(RoleTwo))) ∧
to_be_performed(delete(role_belongs_to_goup(RoleTwo, Group1))) ∧
to_be_performed(delete(intra_group_connection(RoleOne, RoleTwo, Group1, t1))) ∧
to_be_performed(delete(intra_group_connection(RoleTwo, RoleOne, Group1, t1))) ∧
to_be_performed(delete(transfer_property(tp1, RoleOne, RoleTwo, Group1))) ∧
to_be_performed(delete(has_expression(tp1, {expression1}))) ∧
to_be_performed(delete(transfer_property(tp2, RoleTwo, RoleOne, Group1))) ∧
to_be_performed(delete(has_expression(tp2, {expression2})))
The example models the deletion of Role One from Group1. Both specification languages have
been used to model this change as is shown by the braces at the side.
C.3 Building properties for the changing organization
In a change process it is needed that the roles have beliefs about the organization structure.
Therefore all organization structure representations described in Section 4 are included ; some
examples are shown in Table C.2,
Table C.2. Examples of included organization structure representations
exists_role : ROLE → STATPROPEXP
role_belongs_to_group: ROLE x GROUP → STATPROPEXP
role_property: DYNPROP x ROLE x GROUP → STATPROPEXP
has_expression: DYNPROP x DYNPROPEXP → STATPROPEXP
allocated_to: AGENT x ROLE x GROUP → STATPROPEXP
Moreover, to express beliefs, the following language construct is used :
belief: STATPROPEXP → STATPROPEXP
An example of its use is: belief(exists_role(s) ∧ role_belongs_to_group(s, g))
SL
BL
48 Mark Hoogendoorn1, Catholijn M. Jonker2, Martijn C. Schut1, and Jan Treur1
Furthermore it is needed that the roles have beliefs about the behavioral properties that are
expected from a certain role. Therefore first a representation
leads_to: CONSTATOMS x CONSTATOMS → DYNPROPEXP
is introduced for a simple type of such properties. A more general type of dynamic property is
built using:
& : DYNPROPEXP x DYNPROPEXP → DYNPROPEXP
and similarly for other logical connectives such as not, �, ∀, ∃.
Thus within the sort DYNPROPEXP two types of expressions are built:
• temporal statements based on atoms of the form state(γ, t, P) |= p for state properties p
• leads to statements of the form leads_to(V, W) with V and W conjunctions of atoms
Although the latter type of expressions can be mapped to (are definable in terms of) the former
type of expressions, for simplicity they are kept separate.
An example of an expression that can be built using the constructs above is the following
∃t state(γ, t, internal(r)) |= belief(exists_role(s) ∧ role_belongs_to_group(s, g)) ∧
belief(role_property(d1, s, g)) ∧
belief(has_expression(d1, leads_to(a∧b, c)))
This expression states that
there will be a time that
within role r there is the belief that
the organization structure includes role s in group g, and
this role has dynamic property d1 which
is expressed by leads_to(a∧b, c).
Another example property is the following, describing that a role performs the behavior it believes
that is expected from the role:
If at time t
a role believes that
this role has as part of its behavior description that
upon input v the output action w is done,
and
v occurs as input,
then
Modeling Centralized Organization of Organizational Change 49
at a next point in time this role will provide output w.
Here the nesting is visible in the informal structured text representation using tabs. The
formalization of this property also shows a nesting as indicated.
[ state(γ, t, internal(RegHead)) |= belief(role_property(d, RegHead, RegGroup1)) ∧
belief(has_expression(d, leads_to(v, w)))
& state(γ, t, input(RegHead)) |= v ]
� ∃t' ≥ t state(γ, t', output(RegHead)) |= w